|
Abstract:
|
Commonly used phase 1 oncology dose-finding trial designs enroll patients in cohorts and apply sequential decisions that determine the dose level for each cohort based on observed toxicity data. Accrual is suspended after enrollment of each cohort of patients until all the patients in the current cohort have observed outcomes, with or without DLTs. This type of cohort-based designs can be inefficient, especially if the trial needs frequent suspension due to rapid enrollment. This abstract discussing two novel designs to address this issue: 1) Rolling Toxicity Probability Interval Design (R-TPI), a rolling enrollment design that combines the features in mTPI-2; and 2) Time-to-Event Bayesian Optimal Interval (TITE-BOIN) design built upon BOIN. Both designs allow dose escalation/de-escalation rule pre-tabulated prior to trial starts, making it transparent and simple to implement. Numerical studies show that the both designs supports continuous accrual, reduces the length of trial duration, without sacrificing patient safety nor the accuracy of identifying the MTD .
|
